There are a lot of variables that might affect the financial market, and those elements aren't always predictable. As a result, predicting future stock price changes is no easy feat. The goal of machine learning is to discover patterns in massive datasets automatically. Machine learning algorithms have the potential to automate trading methods that are based on their predictions of stock price movements because of their self-organizing and self-learning qualities. Market moves have been predicted using AI methods, yet published methods seldom involve testing in an actual (or virtual) trading setting. In this paper we are in search of a complete automatic trading, where even the buy/sell decisions are taken by the computer.